Social cognitive learning theories have posited a connection between performance and three self-evaluative factors--goal-setting, self- efficacy and control beliefs. This research will begin to explore these relationships, in the domain of memory, as a function of age and varying task conditions. The memory tests will vary from high to low difficulty to observe the impact of challenge on these self- evaluative factors. Three studies will examine two different memory tasks, group vs. individual administration, and varied operationalizations of test difficulty. The hypotheses are comparable across the three studies. In Study 1, high and low difficulty recall tasks starting with 34- or 14-item lists, respectively will be administered for 3 test trials after baseline assessment of recall (with a moderate difficulty list) and memory beliefs. During the 3 test trials, old and young will be assigned to personal goal-setting, goal-setting plus feedback, or control. A final test trial will reassess baseline measures. Hypotheses: goals plus feedback will result in the largest increases between baseline and final trial for performance and beliefs; older adults will show less positive beliefs (low self-efficacy, low perceived control) and poorer performance than the young; age will interact with difficulty such that older adults will show larger differences between high and low difficulty conditions than do younger adults. Study 2 replicates Study 1 with self-paced text recall, with comparable predictions except that goal condition effects are ejected to be stronger under self-paced conditions. Study 3 replicates Study 1 using individualized difficulty levels--the number of items on the high difficulty lists will be an individual's base line score + 20 items and low difficulty lists will be baseline + 4. This procedure of matching difficulty to individual abilities should eliminate age interactions with difficulty level, but other predictions are the same. In all three studies, it is expected that baseline beliefs will predict goals. Regression will be utilized to test predictions concerning the impact of subject variables (e.g., education), baseline beliefs and performance, goal condition and difficulty condition, and age, on final trial beliefs and performance, based on self-efficacy theory (Bandura, 1989b; Schunk,1991).